NVIDIA completes its inference-side closed loop: Partnering with Equinix and Together AI to open up model inference to enterprises.

Nvidia is extending its AI infrastructure footprint from the training side to the inference side.

On Wednesday, NVIDIA partnered with Equinix, the world's largest data center colocation provider, and Together AI, an AI inference platform company, to jointly provide enterprise customers with open model inference services, filling the final gap from model training to inference deployment.

According to the division of labor among the three parties, Equinix provides data center hosting, Together AI provides the inference platform, and NVIDIA provides GPUs and software stack. The three parties package hardware, software, and hosting capabilities to directly enter the inference stage of enterprise-level AI applications.

This collaboration represents a crucial addition to NVIDIA's inference ecosystem. The company has already established a leading position in the training side, and now, by extending its open model inference capabilities to enterprise customers, it further broadens the reach of its computing power ecosystem.

For Equinix, this collaboration also means that it has found a differentiated entry point in the trillion-dollar AI data center construction boom: enterprise-grade open model inference.

Three-way division of labor: a combination of hosting, platform, and computing power.

In this collaboration, Equinix is responsible for data center hosting, Together AI provides the inference platform, and NVIDIA provides the GPUs and software stack.

Amid the trillion-dollar AI data center construction boom, Equinix, as the world's largest data center colocation provider, has positioned itself in the niche market of enterprise-grade open model inference by partnering with NVIDIA and Together AI, becoming a prominent beneficiary in this collaboration.

The core objective of the tripartite collaboration is to help enterprise customers run inference workloads using open models, rather than limiting computing power to a few leading model vendors.

Nvidia's bet on open models is logically sound. CEO Jensen Huang recently published an article elaborating on the importance of open-source AI, which garnered endorsements from almost all major AI companies. In his view, AI models and applications are complementary to Nvidia GPUs, and the flourishing of open-source models means a greater demand for more accurate and widespread computing power.

Inference revenue has surpassed training revenue; Nvidia completes the closed loop.

According to Nvidia's investor conference, about 18 months ago, Nvidia's training and inference revenues were roughly equal; now, inference revenue has surpassed training revenue, and this gap is expected to continue to widen.

At the same time, emerging cloud service providers now contribute more than 50% of AI computing infrastructure revenue, and the growth momentum is spreading from traditional hyperscale cloud vendors to a broader AI computing power ecosystem.

Nvidia has already made numerous moves in the training and architecture sectors.

Groq will bring in its core team and LPU technology through a technology licensing agreement and plans to deeply integrate the LPU into the next-generation Vera Rubin architecture (Groq 3 LPX) ; it also plans to acquire the open-source AI platform Hugging Face for $12.9 billion.

This collaboration with Equinix and Together AI is a key step for NVIDIA to extend its reach to the inference side, enabling its computing ecosystem to cover the entire chain from model training to inference deployment.

Risk Warning and DisclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.